본문 바로가기
  • Home
Basic Search

An Input Video Normalization Method Based on Valid Regions and Frames for Illegal Streaming Video Identification

  • Journal of Software Forensics
  • Abbr : JSF
  • 2026, 22(2), pp.93~105
  • DOI : 10.29056/jsf.2026.06.09
  • Publisher : Korea Software Assessment and Valuation Society
  • Research Area : Engineering > Computer Science
  • Received : May 30, 2026
  • Accepted : June 20, 2026
  • Published : June 30, 2026

Park Byeongchan 1,  Youngmo Kim 1,  UiJin Jang 1

1숭실대학교

Accredited

ABSTRACT

Illegal streaming videos often undergo transformations such as re-encoding, resolution conversion, aspect ratio changes, brightness variation, rotation, and flipping. They may also include non-content regions such as black borders, subtitles, logos, and advertisements during screen recording or retransmission. These factors reduce the consistency of feature vectors extracted from the same content and degrade video identification performance. This paper proposes an input video normalization method based on valid regions and frames for illegal streaming video identification. The proposed method consists of valid region detection, aspect-ratio-preserving resolution normalization, frame refinement based on brightness and structural information, rotation candidate-based orientation correction, and valid frame selection. These steps reduce unnecessary regions and unstable frames, enabling more consistent feature representations in the subsequent feature extraction process. Experimental results using query videos with various distortions show that the proposed method improves identification accuracy compared with the non-normalized condition. It also reduces unnecessary feature vectors while maintaining stable identification performance.

Statistics

  • 16 Viewed
  • 8 Downloaded

Tools

Issue List

Citation status

* References for papers published after 2024 are currently being built.

Basic Search Advanced Search